ABSTRACT
This chapter discusses the role of generalization in adaptive pattern recognition and adaptive motor control processes. It also refers to two types of neurons in the mentioned hierarchy, linear and cyclic nucleotide neurons, and relate their properties to the generalization concept. The evolutionary selection circuits model describes learning as the result of the gradual modification of enzymatic neurons through a process of variation and selection. Enzymatic neurons are formal neurons whose behavior is mediated by excitase enzymes. The underlying principle of the behavior of enzymatic neurons is that excitase enzymes, activated by specific input patterns, cause the neuron to fire. The dynamics of enzymatic neurons operates at two main levels. An upper level describes their behavior in terms of a firing rule that relates the excitase configuration of the neuron to its internal state. Enzymatic neurons are essentially pattern recognizers. Discrete neurons recognize the class of input patterns to which they respond by firing.
